Advanced Micro Devices has increased performance projections for its next-generation code-named Trinity accelerated processing unit and set up launch timeframe for the chip. While AMD Trinity will be faster than initially believed, it will become available only in the middle of the year, which is somewhat later than generally expected based on AMD's comments.
According to performance benchmarks conducted by AMD, the Trinity 35W APU with Piledriver-class x86 cores will provide 25% better x86 performance compared to Llano 35W (with K10.5+ "Husky" x86 cores) based on results obtained in PC Mark Vantage Productivity benchmark. AMD also claims that Trinity 35W will offer up to 50% better result in 3D Mark Vantage performance benchmark compared to Llano 35W.

Earlier released slides, which were also presumably from AMD, projected 20% increase in x86 performance and 30% boost in graphics performance for Trinity compared to currently available A-series "Llano" APUs based on simulations.
Although AMD implied a number of times that it would release Trinity in the first half of the year and rather sooner than later, the official plan now is to launch it in the middle of the year. Some unofficial sources have implied that A-series "Trinity" will be released in June, 2012.
"We are indeed on track with Trinity for mid-2012," said Chris Hook, a spokesman for AMD, without elaborating on actual months or dates.
According to documents seen by X-bit labs, staring from early and middle March, 2012, AMD intends to mass produce its desktop A-series "Trinity" accelerated processing units with 65W thermal design power (TDP). In early May, 2012, the chip designer wants to initiate mass production of desktop A-series "Trinity" APUs with 100W TDP and higher performance.
AMD’s second-generation code-named Trinity APU for mainstream personal computers (Comal for notebooks and Virgo for desktops) will be made using 32nm SOI HKMG process technology at Globalfoundries. The APU will feature up to four x86 cores powered by enhanced Bulldozer/Piledriver architecture, AMD Radeon HD 7000-series "Southern Islands" graphics core with DirectX 11-class graphics support, DDR3 memory controller and other improvements. The chips will be compatible with new infrastructure.
Tags: AMD, Bulldozer, Trinity, Llano, 32nm, Piledriver, Virgo, Comal
Comments currently:
44
Discussion started: 01/18/12 09:10:46 AM
Latest comment: 01/24/12 08:05:06 AM
Expand all threads |
Collapse all threads
[1-10]
1.
For AMD, what you need to understand are the words EXECUTE and DELIVER!
Trinity on track?! These chips are supposed to be released this 1st quarter...
show the post
3
16
[
Posted by: dudde

|
Date: 01/18/12 09:10:46 AM]
+ expand thread (1 answer)
- collapse thread
The Q1 release date was said before Bulldozer's release. This was said sometime in Sept. 2011. Today it looks like late Q2 2012 where we will see Piledriver Desktop Cores based on a modified/treaked design along with the addition of Quad-Channel (Finally). So for now 25% performance over todays Bulldozer is not bad at all. This performance % could easily increase to 30% -50% by the time AMD gets these cores on store shelves. Anything is possible at this time.
With Quad-Channel I can see memory bandwidth going through the roof in performance vs. the current Dual-Channel. Piledriver is looking really good.
3
0
[
Posted by: nt300

|
Date: 01/23/12 06:20:11 PM]
2.
Always read the disclaimers on projected scedules. AMD is subject to the capabilities and reliability of GloFo and TSMC.
17
2
[
Posted by: beenthere

|
Date: 01/18/12 10:43:10 AM]
+ expand thread (9 answers)
- collapse thread
Simply one of your excuses not to deliver or to delay! In any case, delays have always been part of amd's gameplan!
show the post
3
16
[
Posted by: dudde

|
Date: 01/18/12 11:11:05 AM]
delays have always been part of amd's gameplan!
Derp, No company plans for delays, that's just retarded. Delays most likely arise due to things out of the companies control, or random out of the blue issues that no one could have foreseen coming.
3
2
[
Posted by: veli05

|
Date: 01/18/12 11:58:34 AM]
Not in AMD's case, or maybe you need to refresh your memory? Barcelona/Phenom was kinda late... BD was terribly late (years?!?)... it is expected of them to be again late or the right word is again DELAYED!
show the post
3
16
[
Posted by: dudde

|
Date: 01/18/12 12:24:13 PM]
As is Intel, or any other semiconductor company for that matter.
0
2
[
Posted by: veli05

|
Date: 01/18/12 11:56:23 AM]
AMD Trinity is based only on GlobalFoundries with AMD Kaveri possibly is being based on 28nm with GlobalFoundries and STMicroelectrics
AMD Trinity = 32nm-SHP
AMD Kaveri = 28nm-SHP
4
0
[
Posted by: seronx

|
Date: 01/18/12 01:02:00 PM]
You got that right buddy. Its completely insane how there are people still comparing Intel and its fabs to fabless AMD that outsources the production of its designs.
2
1
[
Posted by: bereft

|
Date: 01/18/12 03:34:37 PM]
How come other companies that outsource their production can deliver and market their products on a regular and timely basis.
Just because AMD decided to go the outsourced-route does not excuse AMD from the expectations of their customers and partners to have their product released on a timely manner!
2
2
[
Posted by: PnoyP

|
Date: 01/18/12 03:54:18 PM]
what companies?, the only companies that sell *PUs to consumers are Nvidia (which faces delays), Intel (which has fabs) and AMD
AMD shouldn't have sold it's fabs I agree, but being the only x86 CPU provider with no fabs is hard
3
2
[
Posted by: madooo12

|
Date: 01/18/12 05:47:48 PM]
It needed to sell its fabrication plants in order to reduce its debts. The company may not have survived through the Global Financial Crisis (V 1.0).
14
0
[
Posted by: linuxlowdown

|
Date: 01/19/12 12:20:00 AM]
3.
The haters go from one story to another yet reality does not change. What a horrible life they must lead?
17
5
[
Posted by: beenthere

|
Date: 01/18/12 12:48:30 PM]
+ expand thread (13 answers)
- collapse thread
Like spam artist such as yourself going from one comment to another! wow! pathetic!
show the post
3
16
[
Posted by: dudde

|
Date: 01/18/12 02:24:52 PM]
YES, you are pathetic. Now that you know perhaps you should educate yourself socially and technically and stop the ignorant comments you post?
15
3
[
Posted by: beenthere

|
Date: 01/18/12 04:31:11 PM]
really?!? go screw you mom over your make up stories of amd being the best crap!
show the post
3
15
[
Posted by: dudde

|
Date: 01/18/12 05:06:28 PM]
Trinity is going to suck because can not fix something that is not fixable. The efficiency per core is worst than K10, so the performance of the Bulldozer is equal to Athlon II. Athlon II can not even compete against Intel's i3 processors. That pretty much sums up why Trinity will not be better unless counting on AVX as a reason to upgrade which is a very compulsive decision if going with Trinity as an upgrade.
BTW, you beenthere go from site to site rashing over AMD processors. I like AMD processors too, but smothering people that AMD will make this exciting processor that is going to knock the pixels out the water is just too much for people. It is OK to be a fanboy, but cool it.
4
5
[
Posted by: tecknurd

|
Date: 01/18/12 05:31:12 PM]
maybe AMD made some fixes to bulldozer and made the per-core stuff better, it says it has a 25% better CPU score and a 50% graphics score, if that's true it's worth the wait
we'll have to wait and see
BTW trinity has the same number of cores so the performance gains translate to per-core performance gains
EDIT: i like your thinking, it's OK to be a fan but be cool
1
0
[
Posted by: madooo12

|
Date: 01/18/12 05:40:51 PM]
No, the performance gains are from the increased frequency. Remember that Llano 35W has a clock of 1.5 GHz with turbo to 2.4, which is not really working.
0
1
[
Posted by: cosminmcm

|
Date: 01/18/12 11:04:42 PM]
That was Turbocore 1.0 the Bulldozer review on xbitlabs shows that TurboCore 2.0 is fantastic
but how do you know that they are from frequency increases
0
0
[
Posted by: madooo12

|
Date: 01/19/12 11:19:10 AM]
Well, for start, Bulldozer architecture was made for high frequencies, and as you said the Turbocore will work this time. And the mobile Llano has a very low frequency by itself. The Piledriver cores from Trinity have no L3 cache, and there are only 2 modules compared to 4. What do you think, will they have a higher frequency?
0
0
[
Posted by: cosminmcm

|
Date: 01/20/12 12:25:33 AM]
Nonsense, the performance gains AMD is talking about is Clock for Clock performance vs. current Bulldozer. AMD knows for a FACT they need to make Trinity high performance with low power. To do this they cannot just jack up the Bulldozer’s clock rate and call it Piledriver.
AMD has been working on Enhanced Bulldozer (Piledriver) well before Bulldozer’s release. Bulldozer is a great design in terms of this modular approach. They only need to work/tweak/enhance what they have right now for better clock for clock performance. Also once those new instruction sets get utilized by developers, AMD CPU’s are going to really benefit from them.
3
0
[
Posted by: nt300

|
Date: 01/23/12 06:31:23 PM]
2
0
[
Posted by: nt300

|
Date: 01/23/12 06:37:40 PM]
I disagree with you that Trinity is going to suck (although I am disappointed that Trinity doesn't represent x86 + ARM + AMD graphics... that's still 3 to 5 years away, I suspect). I am not a microprocessor engineer. But I understand that the Bulldozer design allows scaling down wafer sizes that previous architectures could not. That is one of its main design purposes. In scaling down nodes, the processor based on this design will inherently become more efficient. The first generation Bulldozer core at 32nm is indeed not more powerful per core than the previous K10.5 architecture based on 45nm node size. But it allows more of its cores to be put into each wafer - ie the cores each consume less electricity. Design wise, power efficiency is demonstrated to be better in multi-threaded applications (eg video transcoding) but often slightly worse in single thread ones (ie office suites), at the same TDP 125 Watts. But AMD is rightly predicting that multi-threaded application design is the future of performance computing, eventually leading to GPU co-processing. Nothing is broken with this architecture. The first generation Bulldozer was not as sensational as predicted. But its engineering horizon is over at least three nodes and 5 years, with more cores to be implanted (8 - 32).
12
0
[
Posted by: linuxlowdown

|
Date: 01/19/12 01:32:20 AM]
the problem with Zambezi is the low IPC, that's the problem
if they can fix the IPC to be a little better than PII or maybe near Core with the same frequencies the result will be amazing
0
1
[
Posted by: madooo12

|
Date: 01/19/12 11:25:32 AM]
Smart comment. AMD has shown foresight before and their processors often improve with advancement in other components and software and I expect their new architectures to be no different. I think they took the right path and Intel will follow eventually as they have before proving it correct. Hopefully they can make the needed short-term improvements as well.
0
0
[
Posted by: mikato

|
Date: 01/24/12 08:05:06 AM]
4.
The pot calling the kettle black!
I've read most of your post beenthere! really mate?!? you're full of fanboyish crap!
2
4
[
Posted by: PnoyP

|
Date: 01/18/12 05:09:12 PM]
5.
I'll gladly wait the two or three months for the extra performance
we don't want rushed out products like Zambezi anymore do we?
1
0
[
Posted by: madooo12

|
Date: 01/18/12 05:50:12 PM]
6.
Wow, the technical ignorance posted in some of these comments that passes for knowledge is appalling...
Denial ain't a river in Egypt ferchrissake. Some folks here need to start taking their meds again before they jump out a window.
13
4
[
Posted by: beenthere

|
Date: 01/18/12 06:08:31 PM]
+ expand thread (2 answers)
- collapse thread
Wow your fanaticism clearly shows you're in constant drugs! You seem to think that commenting and posting on this site (and others as well) make you the god of nerds around the world.
show the post
3
14
[
Posted by: dudde

|
Date: 01/18/12 07:15:36 PM]
And you think your comments are real?!? Its disgusting!
1
3
[
Posted by: PnoyP

|
Date: 01/18/12 07:18:04 PM]
7.
Yup, some Intel fanatics just are out of touch with reality. It's sad by not uncommon. They simply can't deal with real life.
13
3
[
Posted by: beenthere

|
Date: 01/18/12 07:38:14 PM]
+ expand thread (3 answers)
- collapse thread
and then comes amd fanboys like you with a twisted mind! Oh let me correct myself! mindless retards who think they have life!
show the post
2
15
[
Posted by: dudde

|
Date: 01/18/12 07:41:45 PM]
There are more like you out there, who buy AMD just because they are cheaper. And beause they have AMD, they think they have better processors too.
2
0
[
Posted by: cosminmcm

|
Date: 01/18/12 11:09:03 PM]
You think your the only one with a real life?!? wow what a loser!
0
2
[
Posted by: PnoyP

|
Date: 01/19/12 08:08:34 AM]
8.
southern islands, on chip, with fusion (gpu+cpu sharing memory),... that is going to be one ultra dense compute core!! following very shortly after their southern isles launch, at that: AMD is really pushing SoC hard, and it looks to have the promise of paying off really well.
now if only they'd do something about that nasty oft-unnecessary fusion hub, and add an on die 10Gbit ethernet controller or two. add some on-package memory ala the Radeon E4690 and it'd be everything, right?

what would you be missing?
1
0
[
Posted by: rektide

|
Date: 01/18/12 08:24:30 PM]
9.
According to performance benchmarks conducted by AMD,
Sorry but the slide clearly says it's only the
projection. No words about real benchmarking.
the Trinity 35W APU with Piledriver-class x86 cores will provide 25% better x86 performance compared to Llano 35W (with K10.5+ "Husky" x86 cores)
If true, does it mean that if Zambezi lost L3 cache, the one would have been 25% better?
Is L3 cache unbearable burden for the module architecture?
show the post
1
10
[
Posted by: Azazel

|
Date: 01/18/12 08:36:38 PM]
+ expand thread (1 answer)
- collapse thread
It's not performance per clock. They will run at a higher frequency. It's that simple

0
1
[
Posted by: cosminmcm

|
Date: 01/18/12 11:10:46 PM]
10.
A8-3870(X4) has 3.0GHz.
With +25% it will have 3.75GHz.
Zambezi FX-4100(X4) has 3.6GHz.
You see L3 cache is a burden no matter what angle we look from.
show the post
0
9
[
Posted by: Azazel

|
Date: 01/19/12 08:51:41 AM]
[1-10]